Celeron 366 and 400: the accessible speed
The new Celeron models
The Intel boys have not remained short: neither one, nor two, nor three, but four new Celeron models; good, in fact two models, 366 and 400 MHz, each one with two types different from connector.
Yes, fortunately and almost in contradiction with the first news that were known on the topic, Intel is going to keep on making Celeron for Slot1... for the time being, because there is no doubt that it is going to promote enormously the Socket370, much cheaper, like the future option for the Celeron. But meanwhile, two new mikes for Slot1 do not come badly at all, the truth.
I insist that the models for Slot1 and those of Socket 370 are virtually identical except in his exterior form, although undoubtedly there will exist valuable differences of price between they and even it is possible that it is difficult to obtain models for Slot1, but nothing of this means that they are better or worse; they are simply different.
An exceptional yield
Well, already I know that it seems a little pretentious to begin the part on the yield with a conclusion so impactante, but I am afraid that it is the most absolute reality. I do not believe that anybody accuses me of favor with Intel, in the last years I have used two mikes AMD and a Cyrix with entire satisfaction, but when something is good, since that... to César what belongs to César, although César is the almighty Intel.
The availability of the new Celeron is still limitadísima, especially in Europe, therefore most of the available tests is realized by the proper Intel, but this is not very much a problem since it is possible to compare it with Pentium II of Intel and from this result to compare it with the K6-2.
Sending us to tests that measure the mathematical behavior of the chip and his capacities 3D, like the FPUmark and 3D WinBench 98 of Ziff-Davis, we observe that the differences between mikes Pentium II and Celeron Mendocinos fall down like average below 2 %, what is enclosed inside the possible margin of error of the above mentioned tests. In fact both types of mike have the same nucleus, for what hardly they might give different results in mathematical calculations where the cache memory L2 has little importance and his quantity does not influence by no means, while his speed is enclosed a factor in favor of the Mendocino.
Therefore, these Celeron give good results perfectamenten in mathematical applications, especially in those in which the cache memory quantity has a minor importance: the games. They lack sophistications software as them 3DNow! of the K6-2, but in return they have an impressive brute force at a price much minor than that of a Pentium II. In serious applications as the CAD yes they would be overcome (although not for much) by Pentium II thanks to the biggest cache memory L2 of this one, but: who wants to do CAD when we can play Quake?
More complicated it turns out to be to measure the yield in applications ofimáticas of the new Celeron. If Celeron Mendocino we remember the original report on (available between the Topics Related in the margin), the index iCOMP of Intel was making sure that the Mendocino were, in general lines, 14 % slower than Pentium II at the same speed. This turns out to be very debatable, as we mention already in his moment (between other things, Intel provides Pentium II of discs SCSI for the tests opposite to EIDE for the Celeron); other tests were indicating very different things, as it indicates the following graph:
This is, a difference of little more than 1 % in favor of Pentium II in the test Winstone 98 of Ziff-Davis, that they measure the yield in typical applications like Word, Lotus 123, Netscape Navigator, etc. Consulted diverse opinions in the network, the whole world seems to coincide that the yield of the new Mendocino, although it can turns penalized for having less cache memory than Pentium II, the above mentioned difference patrols concerning 3 %, in any case below 5 %.
So, in general lines, and while there do not appear tests that indicate the opposite (rather they are in favor, for example in some tests realized in www.anandtech.com the difference an insignificant one has left reduced 2 %), Celeron Mendocino to 366 MHz will be at least so rapid (if not more) that Pentium II to 350 MHz, and the model to 400 MHz will stay only a little below his equivalent in Pentium II, but of course: at least of half of price (the mike, not the whole computer, skylight).
As for comparing them with the AMD K6-2, saying that thanks to the fact that his cache memory L2 works at many major speed that 100 MHz of the K6-2, the yield ofimático from the Celeron to 366 MHz will be located minimally over the K6-2 to 350 MHz, while to 400 MHz the Celeron it will be clearly more rapid than the K6-2. It is not for anything that AMD prepares the throwing of the K6-3 with 256 Kb of cache memory integrated at the same speed of the mike, of the form in which it is in the Mendocino...